I'm trying to implement geomorphing on the gpu. My heightmap is stores in a texture and I sample it in the vertex shader. To get the refined height of a vertex, I just sample the heightmap directly. However, I am trying to figure out how to get the height at the next coarser level. Obviously, for a given vertex, it lies on the edge of the coarser grid. So I could find the two edge points and compute the midpoint. In a CPU implementation this seems straightforward to me. But in a vertex shader, this is giving me problems because I do not know which edge of the coarser grid the vertex lies on. When a quad is subdivided, 5 new points are introduced which can lie on a number of edges of the original quad. GPU Gems 2, the article on clip maps, seems to suggest it can be done, but doesn't give details. It just says: "the sample location lies on an edge of the coarser grid, and z_c is obtained by averaging the two course samples on the edge endpoints."

Given the lack of replies here in DX I'll move it over to GP&T - this sounds like something those guys might be able to help you out with [smile]

Cheers,
Jack